US8606815B2 - Systems and methods for analyzing electronic text - Google Patents
Systems and methods for analyzing electronic text Download PDFInfo
- Publication number
- US8606815B2 US8606815B2 US12/331,271 US33127108A US8606815B2 US 8606815 B2 US8606815 B2 US 8606815B2 US 33127108 A US33127108 A US 33127108A US 8606815 B2 US8606815 B2 US 8606815B2
- Authority
- US
- United States
- Prior art keywords
- term
- interest
- electronic text
- sentiment
- taxonomies
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related, expires
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/951—Indexing; Web crawling techniques
Definitions
- Embodiments of the disclosure relate generally to the field of data processing.
- embodiments of the disclosure relate to systems and methods for analyzing electronic text.
- the method includes receiving the electronic text from a plurality of sources.
- the method also includes determining an at least one term of interest to be identified in the electronic text.
- the method further includes identifying a plurality of locations within the electronic text including the at least one term of interest.
- the method also includes for each location within a plurality of locations, creating a snippet from a text segment around the at least one term of interest at the location within the electronic text.
- the method further includes creating multiple taxonomies for the at least one term of interest from the snippets, wherein the taxonomies include an at least one category.
- the method also includes determining co-occurrences between the multiple taxonomies to determine associations between categories of a different taxonomies of the multiple taxonomies.
- FIG. 1 illustrates a system for generating and summarizing taxonomies in analyzing electronic text.
- FIG. 2 illustrates an example method for creating a feature space (dictionary) for the documents by the text clustering module of the system in FIG. 1 .
- FIG. 3 illustrates an example method for indexing documents into categories by the text clustering module of the system in FIG. 1 .
- FIG. 4 illustrates an example method for the time based categorization module of the system in FIG. 1 to create a time based taxonomy.
- FIG. 5 illustrates an example method for the sentiment categorization module of the system in FIG. 1 to score terms in order to score and categorize snippets.
- FIG. 6 illustrates an example method for the sentiment categorization module of the system in FIG. 1 to categorize snippets using the scored terms from the method in FIG. 5 .
- FIG. 7 illustrates an example keywords category vs sentiment category cotable created by the summarization module of the system in FIG. 1 .
- FIG. 8 illustrates an example Dictionary to Category cotable created by the summarization module of the system in FIG. 1 .
- FIG. 9 illustrates an example computer architecture for implementing a system and methods as described in FIGS. 1-6 and the below detailed description.
- Embodiments of the disclosure relate generally to the field of data processing systems.
- embodiments of the disclosure relate to systems and methods for analyzing electronic text.
- numerous specific details are set forth in order to provide a thorough understanding of the present disclosure. It will be apparent, however, to one skilled in the art that the present disclosure may be practiced without some of these specific details. In other instances, well-known structures and devices are shown in block diagram form to avoid obscuring the underlying principles of the present disclosure.
- the system implements an analytical method to discover key brand perception insights without any a priori knowledge.
- the system includes an embedded suite of analytics capabilities to allow brand and reputation analysis.
- the methods for analyzing electronic text include multiple methods for generating useful taxonomies from CGM content and methods to extract perceived brand characteristics.
- Such techniques may be specifically designed for mining blogs and web data with their inherent content quality issues, such as wrong spelling, grammar, etc. However, the techniques are easily adaptable to other better formed content sources, e.g., scientific articles and journals.
- the system extracts insights about a given brand from large corpus of web data.
- large corpus of data human beings often leverage different taxonomies.
- a taxonomy is a structure that groups similar elements of the unstructured information together in a “natural” way and categorized the large document set. Deriving insights from web data can therefore be reduced to the problem of creating groups of similar pieces of the web information together, i.e., multiple taxonomies, in a way that adds value.
- taxonomies In brand perception analysis, correct taxonomies that will help better understand the customer perception of the brand need to be determined and created.
- four taxonomies each with its own specific generation techniques, are effective at uncovering useful brand perception knowledge from web information.
- the four taxonomies include: Text Cluster based taxonomies; time based taxonomies; keyword taxonomies; and sentiment taxonomies. No single taxonomy or the method of generating it alone is sufficient for understand brand perceptions, in and of itself, but the combination of these methods together creates a powerful template for gaining brand insights from web data in a robust and repeatable fashion.
- the four taxonomies are described in detail below.
- FIG. 1 illustrates a system 100 for generating and summarizing taxonomies in analyzing electronic text.
- the system 100 generally comprises a snippet generation module 101 , a text clustering module 102 , a time based categorization module 103 , a keyword categorization module 104 , a sentiment based categorization module 105 , a correlation module 106 , and a summarization module 107 .
- the snippet generation module 101 of the system 100 creates snippets or portions of electronic text for analysis.
- One form of electronic text, web content often is noisy. It may cover diverse topics in one document, even though only a few sentences might be relevant to the analysis subject.
- snippets of electronic text are generated from ingested web data.
- a snippet is defined as a small text segment around a specified keyword.
- the text segment is determined by sentence boundaries.
- the text segment may be determined by number of words, number of characters, or number of sentences.
- the snippet may include the sentence before, after, and comprising a term of interest.
- a term of interest is defined as a term discerning a focused topic and/or brand/company names.
- snippets are built around the terms of interest.
- the disclosure describes converting documents into snippets, but any electronic text may be converted, including web pages, blogs or forum postings.
- the text clustering module 102 of the system 100 creates clusters of documents/electronic text for analysis.
- the user may have no preconceived idea about what categories the document collection should contain, but in other embodiments, the user may start the system with a preliminary grouping of categories.
- text clustering is able to create an initial breakdown of the documents into clusters, by grouping together documents having similar word content.
- documents may be represented in a vector space model.
- each document is represented as a vector of weighted frequencies of the document features.
- Example document features include words, phrases, and strings of characters.
- the text clustering module 102 uses normalized term frequency (t ⁇ n) weighting scheme, which emphasizes words with high frequency in a document, to normalize each document vector to have a unit Euclidean norm (i.e., the magnitude of each vector is 1).
- FIG. 2 illustrates an example method 200 for creating a feature space (dictionary) for the documents by the text clustering module 102 .
- the module 102 determines which words and phrases make up the document feature space by counting which words occur most frequently in the text. In one embodiment, a word occurs most frequently in the text if it appears in the most documents making up the electronic text. Proceeding to 202 , the module 102 removes common words that are not of interest. In one embodiment, the module 102 uses a standard “stop word” list to eliminate common words such as “an”, “a”, “and”, “but”, and “the.”
- the module 102 retains the top N words remaining after removing the stop words.
- the value of N may vary depending on the length of the documents, the number of documents and/or the number of categories to be created.
- the module 102 reviews the electronic text in a second pass that counts the frequency of the two word phrases that occur using the top N words from 203 .
- a two word phrase is defined as two consecutive words having no intervening words that are not stop words.
- the module 102 searches for phrases consisting of more than two words. Proceeding to 205 , the module 102 then prunes the overall list of N frequent words and the two word phrases found in 204 in order to keep the N most frequent words and phrases.
- the N most frequent words and phrases are the feature space (dictionary) of the document corpus.
- the user may edit this feature space (dictionary) as desired to improve clustering performance, such as adding words and phrases the user deems important.
- the module 102 incorporates stemming to create a default synonym table that the user may edit.
- FIG. 3 illustrates an example method 300 for indexing documents into categories by the text clustering module 102 .
- the system 100 implements a document categorization strategy based on categories centered on selected individual terms in the dictionary (keyword taxonomy, as described below).
- the module 102 receives the individual terms of interest.
- the module 102 creates categories from the selected individual terms. Terms of interest may not be known by a user beforehand. Therefore, in one embodiment, the module 102 determines the terms of interest by ranking all discovered terms in the data set based on a normalized measure of cohesion calculated using the formula in equation 1:
- n ⁇ x ⁇ T ⁇ cos ⁇ ( centroid ⁇ ( T ) , x ) ⁇ T ⁇ n ( 1 )
- T is the set of documents that contain a given term
- centroid(T) is the average vector of all these documents
- n is a parameter used to adjust for variance in category size. In one embodiment, n typically equals 0.9.
- the module 102 places the documents into each appropriate category. Upon ranking the terms, in one embodiment, the module 102 selects enough of the most cohesive terms to get 80-90% of the data categorized. Terms may be selected in cohesive order, skipping those terms in the list that do not add a significant number of additional examples (e.g. more than three) to those already categorized with previous terms. In one embodiment, the system 100 halts selection when at least 80% of the data has been categorized and the uncategorized examples are placed in a “Miscellaneous” category.
- the module 102 then employs a single iteration of k-means to the documents within the generated categories to refine the category membership in 304 (i.e. each document is placed in the category of the nearest centroid as calculated by the term membership just described). Hence, each document that contains more than one of the selected terms is placed in the category that is best suited for the overall term content of the document.
- the created clusters are named using the single term that was used to create the cluster, thus avoiding a complex name problem.
- Such a text clustering based taxonomy represents a “table of contents” of the web data around a particular brand or set of brands. It allows the analyst to understand the full spectrum of key words or phrases used that are most commonly used by customers (or the public at large) when discussing the brand.
- the system 100 may create a time based taxonomy using the time based categorization module 103 , a keyword taxonomy using the keyword categorization module 104 , and/or a sentiment based taxonomy using the sentiment based categorization module 105 , as described below.
- Time based taxonomies are classifications of the documents into “chronologically contiguous” categories using the document creation date.
- FIG. 4 illustrates an example method 400 for the time based categorization module 103 of the system 100 to create a time based taxonomy.
- sites of interest which may include, but is not limited to, blogs, message boards and news feeds, are crawled periodically by the module 103 .
- the module 103 may then label each document with a date in 402 . Proceeding to 403 , the module 103 then divides the snippets created by the module 101 from various documents for a given topic into chronologically contiguous categories, wherein the snippets are associated with the date of the document including the snippet.
- Partitions of data based on a calendar may use artificial demarcations including, but not limited to a day, a week, a month and/or a year.
- the appropriate granularity of the partition depends on the span of data being analyzed. In one embodiment, 10-30 categories are sufficient. Therefore, for a year's worth of data, partitions may be by the month. For data spanning a month, the data may be partitioned by days. Calendar based partitions may be useful for drawing trends by a user because the data is easily digestible. Interesting events, though, may span multiple categories and the data may tend to “clump” (be heavily skewed) into one or more categories.
- the module 103 may create equal size categories.
- the module 103 targets a predefined number of categories for indexing the data.
- the number of categories may be independent of the data.
- the module 103 sorts the data chronologically and splits it up into ten approximately equal size categories, each having data that is relatively coincident in time.
- Sample Size based partitions may allow spotting of emerging (or declining) topics over time.
- Each category though, may span a different size time period, thus causing a tougher review for a user.
- the module 103 partitions the data into three primary categories: before, during, and after the event. Such an approach helps determine that the time based partition is related to something important in the data stream. Hence, a user is able to visualize and understand the partitions and data because there are only three classes.
- one taxonomy may be based purely on keywords or terms.
- the keyword categorization module 104 may create a taxonomy from various keywords.
- One example of such a taxonomy is related brands or company names (e.g., brands and companies in a given industry or target similar consumer groups). Often these names are known a prior by a user. In other embodiments, the names may be discovered by perusing the dictionary of terms generated by module 102 or predefined.
- the number of categories in the Brand/Company name taxonomy is usually equal to the number of keywords plus two.
- the two additional categories are: (1) “Miscellaneous” to include snippets that do not mention any brand or company (if any); and (2) “Ambiguous” to include snippets that mention more than one brand/company names.
- An “Ambiguous” grouping allows determining when an event different from discussing a specific brand or company occurs (e.g., to indicate an industry wide issue as opposed to a company/brand specific issue).
- keyword taxonomies can be developed by the keyword categorization module 104 based on a users' domain interest.
- a “hot-issue” based keyword taxonomy may be defined by the keyword categorization module 104 using major customer dissatisfaction keywords, including, but not limited to, slow response, careless service, etc., to create the partitions.
- Keyword taxonomies may capture nearly any issue of a priori importance to the customer, though the keywords themselves may need to be refined to match how the concepts may appear in the data. Occurrence of such “hot-issue” terms in the electronic text is important, regardless of its overall prevalence in the data.
- Sentiment analysis is a technique to gain insight into how a brand or company is perceived.
- a sentiment taxonomy may be partitions of positive sentiment, neutral sentiment, and negative sentiment.
- the sentiment categorization module 105 of the system 100 implements a statistical sentiment analysis approach to measure the relative sentiment on a positive/negative scale expressed by the words in each snippet. The module 105 then generates numeric scores and partitions the snippets into the positive/negative/neutral categories based on the scores.
- FIG. 5 illustrates an example method 500 for the sentiment categorization module 105 of the system 100 to score terms in order to score and categorize snippets.
- the module 105 creates lists of positive and negative words using at least one external Natural Language Processing (NLP) resource.
- NLP Natural Language Processing
- Two example NLP resources include (i) The Inquirer database and (ii) WordNet.
- an internal dictionary may be used.
- the Inquirer database contains more than 4,000 unique words, mostly adjectives. For each word, it defines approximately two hundred Boolean attributes. Some of these attributes are used to decide whether the word is used mostly in a positive sense or in a negative sense.
- WordNet is an online lexical reference system whose design is inspired by current psycholinguistic theories of human lexical memory. English nouns, verbs and adjectives are organized into synonym sets, each representing one underlying concept.
- the module 105 determines for each term in WordNet if most of the synonyms of a word are positive (or negative) according to the Inquirer and then marks the original term as positive (or negative).
- a baseline list of 1905 Positive words and 2282 Negative words was created, which may be used for sentiment scoring.
- the module 105 may receive previously determined results of positive and negative word lists when comparing the two NLP resources (e.g., the 1905 Positive words list and the 2282 Negative words list).
- the module 105 Upon establishing positive and negative sentiment words or terms in 501 , the module 105 establishes the degree of sentiment for terms in 502 . To measure the relative degree of sentiment between different snippets or documents that both use positive/negative words, in one embodiment, the module 105 characterizes the degree (amount) of positive/negative sentiment each sentiment word conveys by looking up the dictionary definition of the word in WordNet and determining a sentiment score for the word as the difference between the occurrence of positive sentiment words and negative sentiment words in the definition. In one embodiment, the occurrence of the word itself in its own definition is counted once if the word appears one or more times, where as other positive/negative words are counted for each occurrence. For further refinement, only adjective or noun definitions may be used and no other part of speech definitions are considered.
- the module 105 normalizes the sentiment scores by dividing the score by the total number of definitions.
- the score indicates the relative sentiment tones each individual word includes. For example, “wonderful” has a score of positive 13, because its one definition includes thirteen positive words.
- the same method for scoring negative/positive sentiment words in the original word lists can be used to score any word in the dictionary generated during text clustering.
- Each word may have both a positive and a negative impact based on having both positive and negative words in its definition.
- the dictionary terms may have less individual impact on the sentiment score than words in the original positive/negative word list.
- words that are not defined in WordNet are ignored for sentiment analysis.
- FIG. 6 illustrates an example method 600 for the sentiment categorization module 105 to categorize snippets using the scored words.
- the module 105 identifies all positive words in a snippet. Proceeding to 602 , the module 105 then sums the positive word sentiment scores for all positive words in the snippet. In 603 , the module 105 identifies all negative words in a snippet. Proceeding to 604 , the module 105 then sums the negative word sentiment scores for all negative words in the snippet. The module 105 then determines the difference between the sum of positive scores and the sum of negative scores in 605 to create a sentiment score for the snippet. The score is then normalized in 606 by dividing by the square root of the snippet length.
- the module 105 then partitions the electronic text by sorting the snippets.
- the module 105 may sort the snippets into quintiles based on the sentiment scores of the snippets.
- the bottom quintile may include the snippets whose sentiment scores are within the lowest 20% of the sentiment scores for all of the snippets in the data.
- the module 105 determines the bottom quintile to be the negative class, the top quintile to be the positive class, and the three middle quintiles to be the neutral class.
- the snippets are sorted into thirds, fourths, or any other portions predetermined by the user.
- the correlation module 106 of the system 100 determines correlations between categories in different taxonomies. For example, a keyword category may have significant overlap of documents or snippets with a time based category. The module 106 may also compare the feature space (dictionary) to the categories and determine any unusually high correlations between terms and categories.
- the primary representation of each category is the centroid of the category.
- the distance metric employed to compare documents or snippets to each other and to the category centroids is the cosine similarity metric, as shown in equation 3:
- documents or snippets when analyzing distance, are not required to belong to the category of its nearest centroid or required to belong to only one category.
- summarization may help in understanding what a category contains without having to analyze all of the documents or snippets in a category.
- the summarization module 107 of the system 100 creates a feature bar chart to help explain what a category contains.
- the chart has an entry for every dictionary term (feature) that occurs in any document of the category.
- Each entry consists of two bars, a first bar to indicate what percentage of the documents in the category contain the feature and a second bar to indicate how frequently the feature occurs in the background population of documents from which the category was drawn.
- the bars are sorted in decreasing order of the difference between first and second bars such that the most important features of a category are shown at the beginning of the chart.
- the chart quickly summarizes the important features of a category, with the feature's relative importance indicated by the size of its bars.
- the summarization module 107 may create other visualization tools, such as pie charts, scatter plots, graphs, etc.
- the module 107 may also sort documents based on the “Most Typical” first or “Least Typical” first criteria. In vector space terms, the module 107 sorts in the order of distance from the category centroid (i.e., most typical is closest to centroid, least typical is furthest from centroid). Reading documents in most typical order may help a user quickly understand what the category is generally about without having to read all documents in the category. Reading the least typical documents can help the user understand the scope of the category and if there is conceptual purity (i.e., do documents exist in the category that should not be related to the category).
- the summarization module 107 may also help determine Category to Category co-occurrences and Dictionary to Category co-occurrences, as described below.
- the summarization module 107 may create a Category to Category co-occurrence to compare different taxonomies to discover where an unusually high degree of association might reside.
- such relationships are visualized via a co-occurrence table (cotable), which shows how the data breaks down across all the combinations of categories in two different taxonomies.
- FIG. 7 illustrates an example keywords category vs sentiment category cotable 700 .
- the challenge with cotables is determining which numbers are interesting. For example, the cell value for Miscellaneous/neutral in cotable 700 is large ( 490 ), thus inferring a possibly important relationship. But, the cell also corresponds to the largest category in the taxonomy. Thus, it should be determined if the number in the cell is more than would have been expected instead of relying on the pure size of numbers in the cells.
- the summarization module 107 determines mathematically if the number is approximately what is expected. Assuming no relationship exists that ties a particular sentiment to particular brand, it is expected to find an expected value E approximately equal to X*Y in a cell for a given sentiment and brand, where X is the percentage of times a given sentiment occurs and Y is the percentage of times the brand occurs. An exceptional value is something greater than E, indicating an association stronger than expected. E is different for every cell in the cotable, since the number of snippets for each sentiment and brand differs.
- the module 107 may get a relative significance for different values in the cotable, such as distinguishing whether a 5 occurring in one cell is more interesting than a 10 occurring in a different cell, by using the Chi-Squared test—a statistical test that calculates the likelihood of seeing any particular value in the cell of a cotable. The smaller this probability, the less likely the value, the more interesting it is from a data mining perspective. When a very low probability value occurs in a cell, it suggests that the original assumption about no relationship existing between the taxonomies was incorrect. A correlation does not signify a definitive relationship between categories, but it may help reveal a relationship by indicating an area for further investigation.
- the module 107 may shade the cells of the cotable according to show calculated probabilities for the values in the cells.
- the module 107 analyzes another type of co-occurrence called Dictionary to Category Co-occurrence by comparing a taxonomy of snippets or documents to the dictionary of words created during text clustering by the text clustering module 102 .
- a Dictionary to Category cotable contains the columns of taxonomy categories and the rows of dictionary terms, sorted in alphabetical order.
- FIG. 8 illustrates an example Dictionary to Category cotable 800 .
- categories may include time, keywords, and sentiment.
- Text clustering may reveal significantly large topics and themes driving opinion on the web, but it may miss smaller, less prevalent issues that arise only for a particular brand or in a small window of time. Thus, Dictionary co-occurrence fills in such a gap.
- FIG. 9 illustrates an example computer architecture for implementing a system and methods as described in FIGS. 1-6 and in the above disclosure.
- the exemplary computing system of FIG. 9 includes: 1) one or more processors 901 ; 2) a memory control hub (MCH) 902 ; 3) a system memory 903 (of which different types exist such as DDR RAM, EDO RAM, etc,); 4) a cache 904 ; 5) an I/O control hub (ICH) 905 ; 6) a graphics processor 906 ; 7) a display/screen 907 (of which different types exist such as Cathode Ray Tube (CRT), Thin Film Transistor (TFT), Liquid Crystal Display (LCD), DPL, etc.); and/or 8) one or more I/O devices 908 .
- CTR Cathode Ray Tube
- TFT Thin Film Transistor
- LCD Liquid Crystal Display
- the one or more processors 901 execute instructions in order to perform whatever software routines the computing system implements.
- the instructions frequently involve some sort of operation performed upon data.
- Both data and instructions are stored in system memory 903 and cache 904 .
- Cache 904 is typically designed to have shorter latency times than system memory 903 .
- cache 904 might be integrated onto the same silicon chip(s) as the processor(s) and/or constructed with faster SRAM cells whilst system memory 903 might be constructed with slower DRAM cells.
- System memory 903 is deliberately made available to other components within the computing system.
- the data received from various interfaces to the computing system e.g., keyboard and mouse, printer port, LAN port, modem port, etc.
- an internal storage element of the computing system e.g., hard disk drive
- system memory 903 prior to their being operated upon by the one or more processor(s) 901 in the implementation of a software program.
- data that a software program determines should be sent from the computing system to an outside entity through one of the computing system interfaces, or stored into an internal storage element is often temporarily queued in system memory 903 prior to its being transmitted or stored.
- the ICH 905 is responsible for ensuring that such data is properly passed between the system memory 903 and its appropriate corresponding computing system interface (and internal storage device if the computing system is so designed).
- the MCH 902 is responsible for managing the various contending requests for system memory 903 access amongst the processor(s) 901 , interfaces and internal storage elements that may proximately arise in time with respect to one another.
- I/O devices 908 are also implemented in a typical computing system. I/O devices generally are responsible for transferring data to and/or from the computing system (e.g., a networking adapter); or, for large scale non-volatile storage within the computing system (e.g., hard disk drive).
- ICH 905 has bi-directional point-to-point links between itself and the observed I/O devices 908 .
- modules of the different embodiments of the described system may include software, hardware, firmware, or any combination thereof.
- the modules may be software programs available to the public or special or general purpose processors running proprietary or public software.
- the software may also be specialized programs written specifically for signature creation and organization and recompilation management.
- storage of the system may include, but is not limited to, hardware (such as floppy diskettes, optical disks, CD-ROMs, and magneto-optical disks, ROMs, RAMs, EPROMs, EEPROMs, flash, magnetic or optical cards, propagation media or other type of media/machine-readable medium), software (such as instructions to require storage of information on a hardware storage unit, or any combination thereof.
- elements of the present invention may also be provided as a machine-readable medium for storing the machine-executable instructions.
- the machine-readable medium may include, but is not limited to, floppy diskettes, optical disks, CD-ROMs, and magneto-optical disks, ROMs, RAMs, EPROMs, EEPROMs, flash, magnetic or optical cards, propagation media or other type of media/machine-readable medium suitable for storing electronic instructions.
- embodiments of the invention may include the various processes as set forth above.
- the processes may be embodied in machine-executable instructions which cause a general-purpose or special-purpose processor to perform certain steps.
- these processes may be performed by specific hardware components that contain hardwired logic for performing the processes, or by any combination of programmed computer components and custom hardware components.
- Embodiments of the invention do not require all of the various processes presented, and it may be conceived by one skilled in the art as to how to practice the embodiments of the invention without specific processes presented or with extra processes not presented.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
Description
wherein T is the set of documents that contain a given term, centroid(T) is the average vector of all these documents, and n is a parameter used to adjust for variance in category size. In one embodiment, n typically equals 0.9.
Terms that score relatively high from
-
- 1. What issues are emerging as important relatively recently in the data (time vs. dictionary)?
- 2. What issues are associated with a particular company or brand (keyword vs dictionary)?
- 3. What issues are driving negative/positive opinions for a given brand (sentiment vs. dictionary)?
Claims (16)
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US12/331,271 US8606815B2 (en) | 2008-12-09 | 2008-12-09 | Systems and methods for analyzing electronic text |
TW098140575A TW201033823A (en) | 2008-12-09 | 2009-11-27 | Systems and methods for analyzing electronic text |
PCT/EP2009/066222 WO2010066616A1 (en) | 2008-12-09 | 2009-12-02 | Systems and methods for analyzing electronic text |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US12/331,271 US8606815B2 (en) | 2008-12-09 | 2008-12-09 | Systems and methods for analyzing electronic text |
Publications (2)
Publication Number | Publication Date |
---|---|
US20100145940A1 US20100145940A1 (en) | 2010-06-10 |
US8606815B2 true US8606815B2 (en) | 2013-12-10 |
Family
ID=42094148
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US12/331,271 Expired - Fee Related US8606815B2 (en) | 2008-12-09 | 2008-12-09 | Systems and methods for analyzing electronic text |
Country Status (3)
Country | Link |
---|---|
US (1) | US8606815B2 (en) |
TW (1) | TW201033823A (en) |
WO (1) | WO2010066616A1 (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20120204104A1 (en) * | 2009-10-11 | 2012-08-09 | Patrick Sander Walsh | Method and system for document presentation and analysis |
US9477704B1 (en) * | 2012-12-31 | 2016-10-25 | Teradata Us, Inc. | Sentiment expression analysis based on keyword hierarchy |
US20170004128A1 (en) * | 2015-07-01 | 2017-01-05 | Institute for Sustainable Development | Device and method for analyzing reputation for objects by data mining |
US20200193380A1 (en) * | 2018-12-14 | 2020-06-18 | International Business Machines Corporation | Detection and categorization of electronic communications |
US11205044B1 (en) | 2009-11-03 | 2021-12-21 | Alphasense OY | User interface for use with a search engine for searching financial related documents |
US11205103B2 (en) | 2016-12-09 | 2021-12-21 | The Research Foundation for the State University | Semisupervised autoencoder for sentiment analysis |
US11651016B2 (en) | 2018-08-09 | 2023-05-16 | Walmart Apollo, Llc | System and method for electronic text classification |
Families Citing this family (46)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7752043B2 (en) | 2006-09-29 | 2010-07-06 | Verint Americas Inc. | Multi-pass speech analytics |
US20160217488A1 (en) * | 2007-05-07 | 2016-07-28 | Miles Ward | Systems and methods for consumer-generated media reputation management |
US8019742B1 (en) | 2007-05-31 | 2011-09-13 | Google Inc. | Identifying related queries |
US8594996B2 (en) | 2007-10-17 | 2013-11-26 | Evri Inc. | NLP-based entity recognition and disambiguation |
US8700604B2 (en) | 2007-10-17 | 2014-04-15 | Evri, Inc. | NLP-based content recommender |
US9183323B1 (en) | 2008-06-27 | 2015-11-10 | Google Inc. | Suggesting alternative query phrases in query results |
US8719016B1 (en) | 2009-04-07 | 2014-05-06 | Verint Americas Inc. | Speech analytics system and system and method for determining structured speech |
US8533208B2 (en) | 2009-09-28 | 2013-09-10 | Ebay Inc. | System and method for topic extraction and opinion mining |
US8620906B2 (en) * | 2009-11-06 | 2013-12-31 | Ebay Inc. | Detecting competitive product reviews |
US8356025B2 (en) * | 2009-12-09 | 2013-01-15 | International Business Machines Corporation | Systems and methods for detecting sentiment-based topics |
US8849785B1 (en) * | 2010-01-15 | 2014-09-30 | Google Inc. | Search query reformulation using result term occurrence count |
US8620849B2 (en) * | 2010-03-10 | 2013-12-31 | Lockheed Martin Corporation | Systems and methods for facilitating open source intelligence gathering |
US8645125B2 (en) | 2010-03-30 | 2014-02-04 | Evri, Inc. | NLP-based systems and methods for providing quotations |
US8788260B2 (en) * | 2010-05-11 | 2014-07-22 | Microsoft Corporation | Generating snippets based on content features |
US8838633B2 (en) * | 2010-08-11 | 2014-09-16 | Vcvc Iii Llc | NLP-based sentiment analysis |
US8725739B2 (en) | 2010-11-01 | 2014-05-13 | Evri, Inc. | Category-based content recommendation |
US8423551B1 (en) * | 2010-11-05 | 2013-04-16 | Google Inc. | Clustering internet resources |
US8370328B2 (en) * | 2011-01-31 | 2013-02-05 | Comsort, Inc. | System and method for creating and maintaining a database of disambiguated entity mentions and relations from a corpus of electronic documents |
US10445677B2 (en) | 2011-03-28 | 2019-10-15 | International Business Machines Corporation | System and method for integrating text analytics driven social metrics into business architecture |
US8650198B2 (en) | 2011-08-15 | 2014-02-11 | Lockheed Martin Corporation | Systems and methods for facilitating the gathering of open source intelligence |
WO2013033385A1 (en) * | 2011-08-30 | 2013-03-07 | E-Rewards, Inc. | System and method for generating a knowledge metric using qualitative internet data |
US9152625B2 (en) * | 2011-11-14 | 2015-10-06 | Microsoft Technology Licensing, Llc | Microblog summarization |
US8812527B2 (en) | 2011-11-29 | 2014-08-19 | International Business Machines Corporation | Automatically recommending asynchronous discussion forum posts during a real-time collaboration |
US9002848B1 (en) | 2011-12-27 | 2015-04-07 | Google Inc. | Automatic incremental labeling of document clusters |
US20130173254A1 (en) * | 2011-12-31 | 2013-07-04 | Farrokh Alemi | Sentiment Analyzer |
US9613323B2 (en) * | 2012-01-05 | 2017-04-04 | International Business Machines Corporation | Organizational agility determination across multiple computing domains |
US9336205B2 (en) * | 2012-04-10 | 2016-05-10 | Theysay Limited | System and method for analysing natural language |
WO2013170345A1 (en) * | 2012-05-15 | 2013-11-21 | Whyz Technologies Limited | Method and system relating to re-labelling multi-document clusters |
US9760592B2 (en) | 2014-02-20 | 2017-09-12 | International Business Machines Corporation | Metrics management and monitoring system for service transition and delivery management |
US10122666B2 (en) * | 2014-03-11 | 2018-11-06 | International Business Machines Corporation | Retrieving and reusing stored message content |
US20150324481A1 (en) * | 2014-05-06 | 2015-11-12 | International Business Machines Corporation | Building Entity Relationship Networks from n-ary Relative Neighborhood Trees |
CN106796583A (en) * | 2014-07-07 | 2017-05-31 | 机械地带有限公司 | System and method for recognizing and advising emoticon |
CN104679728B (en) * | 2015-02-06 | 2018-08-31 | 中国农业大学 | A kind of text similarity detection method |
TWI550422B (en) * | 2015-04-08 | 2016-09-21 | 雲拓科技有限公司 | Claim text generalizing method |
TWI639927B (en) | 2016-05-27 | 2018-11-01 | 雲拓科技有限公司 | Method for corresponding element symbols in the specification to the corresponding element terms in claims |
CN107767195A (en) * | 2016-08-16 | 2018-03-06 | 阿里巴巴集团控股有限公司 | The display systems and displaying of description information, generation method and electronic equipment |
TWI598751B (en) | 2016-12-05 | 2017-09-11 | 雲拓科技有限公司 | Automatic claim computerized-translating apparatus |
US10831796B2 (en) * | 2017-01-15 | 2020-11-10 | International Business Machines Corporation | Tone optimization for digital content |
TWI659321B (en) * | 2018-01-19 | 2019-05-11 | Yuan Ze University | System and method for analyzing industry relevance |
US11238519B1 (en) * | 2018-08-01 | 2022-02-01 | Amazon Technologies, Inc. | Search results based on user-submitted content |
US10713329B2 (en) * | 2018-10-30 | 2020-07-14 | Longsand Limited | Deriving links to online resources based on implicit references |
US11487936B2 (en) * | 2020-05-27 | 2022-11-01 | Capital One Services, Llc | System and method for electronic text analysis and contextual feedback |
TWI793432B (en) * | 2020-08-07 | 2023-02-21 | 國立中央大學 | Document management method and system for engineering project |
CN113010669B (en) * | 2020-12-24 | 2022-06-21 | 华戎信息产业有限公司 | News classification method and system |
US11966427B2 (en) * | 2021-07-07 | 2024-04-23 | Volvo Car Corporation | Keyword-object taxonomy generation and utilization |
CN113642323B (en) * | 2021-08-19 | 2023-09-22 | 成都理工大学 | Method, medium and equipment for detecting evolution trend of research hot spot based on network structure |
Citations (28)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20030046080A1 (en) | 1998-10-09 | 2003-03-06 | Donald J. Hejna | Method and apparatus to determine and use audience affinity and aptitude |
US20030115188A1 (en) * | 2001-12-19 | 2003-06-19 | Narayan Srinivasa | Method and apparatus for electronically extracting application specific multidimensional information from a library of searchable documents and for providing the application specific information to a user application |
US20030182123A1 (en) * | 2000-09-13 | 2003-09-25 | Shunji Mitsuyoshi | Emotion recognizing method, sensibility creating method, device, and software |
US20050086592A1 (en) * | 2003-10-15 | 2005-04-21 | Livia Polanyi | Systems and methods for hybrid text summarization |
US20050091038A1 (en) * | 2003-10-22 | 2005-04-28 | Jeonghee Yi | Method and system for extracting opinions from text documents |
US20050165819A1 (en) * | 2004-01-14 | 2005-07-28 | Yoshimitsu Kudoh | Document tabulation method and apparatus and medium for storing computer program therefor |
US20050203900A1 (en) * | 2004-03-08 | 2005-09-15 | Shogakukan, Inc. | Associative retrieval system and associative retrieval method |
US6952700B2 (en) | 2001-03-22 | 2005-10-04 | International Business Machines Corporation | Feature weighting in κ-means clustering |
US20050283680A1 (en) * | 2004-06-21 | 2005-12-22 | Fujitsu Limited | Apparatus, method, and computer product for pattern detection |
US20060069589A1 (en) * | 2004-09-30 | 2006-03-30 | Nigam Kamal P | Topical sentiments in electronically stored communications |
US20060200342A1 (en) * | 2005-03-01 | 2006-09-07 | Microsoft Corporation | System for processing sentiment-bearing text |
US20060235843A1 (en) | 2005-01-31 | 2006-10-19 | Textdigger, Inc. | Method and system for semantic search and retrieval of electronic documents |
US20060242190A1 (en) * | 2005-04-26 | 2006-10-26 | Content Analyst Comapny, Llc | Latent semantic taxonomy generation |
US20070016580A1 (en) * | 2005-07-15 | 2007-01-18 | International Business Machines Corporation | Extracting information about references to entities rom a plurality of electronic documents |
US20070033221A1 (en) * | 1999-06-15 | 2007-02-08 | Knova Software Inc. | System and method for implementing a knowledge management system |
US20070073745A1 (en) * | 2005-09-23 | 2007-03-29 | Applied Linguistics, Llc | Similarity metric for semantic profiling |
US20070073758A1 (en) | 2005-09-23 | 2007-03-29 | Redcarpet, Inc. | Method and system for identifying targeted data on a web page |
US20070094600A1 (en) * | 2005-10-24 | 2007-04-26 | International Business Machines Corporation | Automatic 3d object generation and deformation for representation of data files based on taxonomy classification |
EP1818839A1 (en) | 2006-02-14 | 2007-08-15 | Accenture Global Services GmbH | System and method for online information analysis |
US20070208569A1 (en) * | 2006-03-03 | 2007-09-06 | Balan Subramanian | Communicating across voice and text channels with emotion preservation |
US20070294281A1 (en) | 2006-05-05 | 2007-12-20 | Miles Ward | Systems and methods for consumer-generated media reputation management |
US20070294230A1 (en) | 2006-05-31 | 2007-12-20 | Joshua Sinel | Dynamic content analysis of collected online discussions |
US20080114755A1 (en) | 2006-11-15 | 2008-05-15 | Collective Intellect, Inc. | Identifying sources of media content having a high likelihood of producing on-topic content |
US20080195567A1 (en) | 2007-02-13 | 2008-08-14 | International Business Machines Corporation | Information mining using domain specific conceptual structures |
US20080215607A1 (en) | 2007-03-02 | 2008-09-04 | Umbria, Inc. | Tribe or group-based analysis of social media including generating intelligence from a tribe's weblogs or blogs |
US7689557B2 (en) * | 2005-06-07 | 2010-03-30 | Madan Pandit | System and method of textual information analytics |
US7774360B2 (en) * | 2006-09-08 | 2010-08-10 | Microsoft Corporation | Building bridges for web query classification |
US8417713B1 (en) * | 2007-12-05 | 2013-04-09 | Google Inc. | Sentiment detection as a ranking signal for reviewable entities |
-
2008
- 2008-12-09 US US12/331,271 patent/US8606815B2/en not_active Expired - Fee Related
-
2009
- 2009-11-27 TW TW098140575A patent/TW201033823A/en unknown
- 2009-12-02 WO PCT/EP2009/066222 patent/WO2010066616A1/en active Application Filing
Patent Citations (31)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7043433B2 (en) | 1998-10-09 | 2006-05-09 | Enounce, Inc. | Method and apparatus to determine and use audience affinity and aptitude |
US20030046080A1 (en) | 1998-10-09 | 2003-03-06 | Donald J. Hejna | Method and apparatus to determine and use audience affinity and aptitude |
US20070033221A1 (en) * | 1999-06-15 | 2007-02-08 | Knova Software Inc. | System and method for implementing a knowledge management system |
US20030182123A1 (en) * | 2000-09-13 | 2003-09-25 | Shunji Mitsuyoshi | Emotion recognizing method, sensibility creating method, device, and software |
US6952700B2 (en) | 2001-03-22 | 2005-10-04 | International Business Machines Corporation | Feature weighting in κ-means clustering |
US20030115188A1 (en) * | 2001-12-19 | 2003-06-19 | Narayan Srinivasa | Method and apparatus for electronically extracting application specific multidimensional information from a library of searchable documents and for providing the application specific information to a user application |
US20050086592A1 (en) * | 2003-10-15 | 2005-04-21 | Livia Polanyi | Systems and methods for hybrid text summarization |
US20050091038A1 (en) * | 2003-10-22 | 2005-04-28 | Jeonghee Yi | Method and system for extracting opinions from text documents |
US20050165819A1 (en) * | 2004-01-14 | 2005-07-28 | Yoshimitsu Kudoh | Document tabulation method and apparatus and medium for storing computer program therefor |
US20050203900A1 (en) * | 2004-03-08 | 2005-09-15 | Shogakukan, Inc. | Associative retrieval system and associative retrieval method |
US20050283680A1 (en) * | 2004-06-21 | 2005-12-22 | Fujitsu Limited | Apparatus, method, and computer product for pattern detection |
US20060069589A1 (en) * | 2004-09-30 | 2006-03-30 | Nigam Kamal P | Topical sentiments in electronically stored communications |
US7523085B2 (en) | 2004-09-30 | 2009-04-21 | Buzzmetrics, Ltd An Israel Corporation | Topical sentiments in electronically stored communications |
US20060235843A1 (en) | 2005-01-31 | 2006-10-19 | Textdigger, Inc. | Method and system for semantic search and retrieval of electronic documents |
US20060200342A1 (en) * | 2005-03-01 | 2006-09-07 | Microsoft Corporation | System for processing sentiment-bearing text |
US20060242190A1 (en) * | 2005-04-26 | 2006-10-26 | Content Analyst Comapny, Llc | Latent semantic taxonomy generation |
US7689557B2 (en) * | 2005-06-07 | 2010-03-30 | Madan Pandit | System and method of textual information analytics |
US20070016580A1 (en) * | 2005-07-15 | 2007-01-18 | International Business Machines Corporation | Extracting information about references to entities rom a plurality of electronic documents |
US20070073758A1 (en) | 2005-09-23 | 2007-03-29 | Redcarpet, Inc. | Method and system for identifying targeted data on a web page |
US20070073745A1 (en) * | 2005-09-23 | 2007-03-29 | Applied Linguistics, Llc | Similarity metric for semantic profiling |
US20070094600A1 (en) * | 2005-10-24 | 2007-04-26 | International Business Machines Corporation | Automatic 3d object generation and deformation for representation of data files based on taxonomy classification |
EP1818839A1 (en) | 2006-02-14 | 2007-08-15 | Accenture Global Services GmbH | System and method for online information analysis |
US20070208569A1 (en) * | 2006-03-03 | 2007-09-06 | Balan Subramanian | Communicating across voice and text channels with emotion preservation |
US20070294281A1 (en) | 2006-05-05 | 2007-12-20 | Miles Ward | Systems and methods for consumer-generated media reputation management |
US7720835B2 (en) | 2006-05-05 | 2010-05-18 | Visible Technologies Llc | Systems and methods for consumer-generated media reputation management |
US20070294230A1 (en) | 2006-05-31 | 2007-12-20 | Joshua Sinel | Dynamic content analysis of collected online discussions |
US7774360B2 (en) * | 2006-09-08 | 2010-08-10 | Microsoft Corporation | Building bridges for web query classification |
US20080114755A1 (en) | 2006-11-15 | 2008-05-15 | Collective Intellect, Inc. | Identifying sources of media content having a high likelihood of producing on-topic content |
US20080195567A1 (en) | 2007-02-13 | 2008-08-14 | International Business Machines Corporation | Information mining using domain specific conceptual structures |
US20080215607A1 (en) | 2007-03-02 | 2008-09-04 | Umbria, Inc. | Tribe or group-based analysis of social media including generating intelligence from a tribe's weblogs or blogs |
US8417713B1 (en) * | 2007-12-05 | 2013-04-09 | Google Inc. | Sentiment detection as a ranking signal for reviewable entities |
Non-Patent Citations (20)
Title |
---|
Amit Behal et al., "Business Insights Workbench-An Interactive Insights Discovery Solution", Proc. of the 12th Int'l Conf. on Human-Computer Interaction, 2007, pp. 834-843. |
Bin He et al., "BIwTL: A Business Information Warehouse Toolkit and Language for Warehousing Simplification and Automation", Proc. of the ACM SIGMOD '07, Jun. 11-14, 2007, Beijing, China, pp. 1041-1052. |
D. Modha et al., "Feature Weighting in K-Means Clustering", Machine Learning, vol. 52, No. 3, 2003, pp. 217-237. |
E. Rich et al., "Artificial Intelligence, 2nd Edition", Publisher: McGraw-Hill Inc. 1991. Chapter 16, Section 16.3 "Distributed Reasoning Systems", pp. 433-444. |
IBM, Ascential. Available at http://www.ibm.ascential.com, 5 Pages as printed from the web Oct. 7, 2010. |
IBM, DB2 Data Warehouse Edition. Available at http://www-306.ibm.com/software/data/db2/dwe, 3 pages as printed from the web on Oct. 8, 2010. |
International Search Report and written opinion dated May 6, 2010, for International Appl. No. PCT/EP2009/066222. |
J. Han et el., "Data Mining: Concept and Techniques", Publisher: Morgan Kaufman, Academic Press 2001. Chapter 2, Section 2.2.2. "Stars, Snowflakes, and Fact Constellations: Schemas for Multidimensional Databases", pp. 48-50, Section 2.2.3 "Examples for Defining Star, Snowflake, and Fact Constellation Schemas", pp. 52-53. Chapter 9, Mining Complex Types of Data, "Example 9.5-9.6", pp. 406-408. |
Java Tutorials Essential Classes. Available at http://java.sun.com/docs/books/tutorial/essential/regex/, 2 pages printed from the web on Oct. 8, 2010. |
Kalido, Enterprise Data Warehousing. Available at http://kalido.com, 2 pages as printed from the web on Nov. 3, 2010. |
Lucene. Available at http://lucene.apache.org, 9 pages as printed from the web on Oct. 8, 2010. |
Mete Celik Shashi Shekhar James P. Rogers James A. Shine Jin Soung Yoo "Mixed-Drove Spatio-Temporal Co-occurrence Pattern Mining: A Summary of Results", 2006, IEEE. * |
R. Srikant et al., "Mining Association Rules with Item Constraints", Proc. of 3rd Int'l Conf. on Knowledge Discovery in Databases and Data Mining, Aug. 1997, Newport Beach, CA., pp. 67-73. |
Robert J. Bayardo et al., "Mining the Most Interesting Rules", Proc. of the Fifth ACM SIGKDD Int'l. Conf. on Knowledge Discovery and Data Mining, 1999, pp. 145-154. |
S. Spangler et al., "COBRA-Mining Web for COrporate Brand and Reputation Analysis", Proc. of IEEE/WIC/ACM International Conference on Web Intelligence, 2007, pp. 11-17. |
S. Spangler et al., "Generating and Browsing Multiple Taxonomies Over a Document Collection", Journal of Management Information Systems, vol. 19, No. 4, 2003, pp. 191-212. |
S. Spangler et al., "Machines in the conversation: Detecting themes and trends in informal communication streams", IBM Systems Journal, vol. 45, No. 4, 2006, p. 785. |
S. Spangler et al., "MindMap: Utilizing Multiple Taxonomies and Visualization to Understand a Document Collection", Proceedings of the 35th Hawaii Int'l. Conference on System Sciences, (HICSS-35'02), pp. 1170-1179. |
S. Spangler et al., "Multi-Taxonomy: Determining Perceived Brand Characteristics from Web Data", Proc. of WI-IAT '08, IEEE/WIC/ACM International Conf. on Web Intelligence and Intelligent Agent Technology, 2008, pp. 258-264. |
W. S. Spangler J. T. KreulenJ. F. Newswanger "Machines in the conversation: Detecting themes and trends in informal communication streams" IBM Systems Journal, vol. 45, No. 4, 2006. * |
Cited By (30)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20120204104A1 (en) * | 2009-10-11 | 2012-08-09 | Patrick Sander Walsh | Method and system for document presentation and analysis |
US8739032B2 (en) * | 2009-10-11 | 2014-05-27 | Patrick Sander Walsh | Method and system for document presentation and analysis |
US20140229475A1 (en) * | 2009-10-11 | 2014-08-14 | Patrick Sander Walsh | Method and system for document presentation and analysis |
US11699036B1 (en) | 2009-11-03 | 2023-07-11 | Alphasense OY | User interface for use with a search engine for searching financial related documents |
US11474676B1 (en) | 2009-11-03 | 2022-10-18 | Alphasense OY | User interface for use with a search engine for searching financial related documents |
US11989510B1 (en) | 2009-11-03 | 2024-05-21 | Alphasense OY | User interface for use with a search engine for searching financial related documents |
US11972207B1 (en) | 2009-11-03 | 2024-04-30 | Alphasense OY | User interface for use with a search engine for searching financial related documents |
US11907510B1 (en) | 2009-11-03 | 2024-02-20 | Alphasense OY | User interface for use with a search engine for searching financial related documents |
US11205044B1 (en) | 2009-11-03 | 2021-12-21 | Alphasense OY | User interface for use with a search engine for searching financial related documents |
US11907511B1 (en) | 2009-11-03 | 2024-02-20 | Alphasense OY | User interface for use with a search engine for searching financial related documents |
US11216164B1 (en) | 2009-11-03 | 2022-01-04 | Alphasense OY | Server with associated remote display having improved ornamentality and user friendliness for searching documents associated with publicly traded companies |
US11227109B1 (en) * | 2009-11-03 | 2022-01-18 | Alphasense OY | User interface for use with a search engine for searching financial related documents |
US11244273B1 (en) | 2009-11-03 | 2022-02-08 | Alphasense OY | System for searching and analyzing documents in the financial industry |
US11861148B1 (en) | 2009-11-03 | 2024-01-02 | Alphasense OY | User interface for use with a search engine for searching financial related documents |
US11281739B1 (en) | 2009-11-03 | 2022-03-22 | Alphasense OY | Computer with enhanced file and document review capabilities |
US11347383B1 (en) | 2009-11-03 | 2022-05-31 | Alphasense OY | User interface for use with a search engine for searching financial related documents |
US11809691B1 (en) | 2009-11-03 | 2023-11-07 | Alphasense OY | User interface for use with a search engine for searching financial related documents |
US11550453B1 (en) | 2009-11-03 | 2023-01-10 | Alphasense OY | User interface for use with a search engine for searching financial related documents |
US11561682B1 (en) | 2009-11-03 | 2023-01-24 | Alphasense OY | User interface for use with a search engine for searching financial related documents |
US11740770B1 (en) | 2009-11-03 | 2023-08-29 | Alphasense OY | User interface for use with a search engine for searching financial related documents |
US11687218B1 (en) | 2009-11-03 | 2023-06-27 | Alphasense OY | User interface for use with a search engine for searching financial related documents |
US11704006B1 (en) | 2009-11-03 | 2023-07-18 | Alphasense OY | User interface for use with a search engine for searching financial related documents |
US9477704B1 (en) * | 2012-12-31 | 2016-10-25 | Teradata Us, Inc. | Sentiment expression analysis based on keyword hierarchy |
US20170004128A1 (en) * | 2015-07-01 | 2017-01-05 | Institute for Sustainable Development | Device and method for analyzing reputation for objects by data mining |
US9990356B2 (en) * | 2015-07-01 | 2018-06-05 | Institute of Sustainable Development | Device and method for analyzing reputation for objects by data mining |
US11205103B2 (en) | 2016-12-09 | 2021-12-21 | The Research Foundation for the State University | Semisupervised autoencoder for sentiment analysis |
US11651016B2 (en) | 2018-08-09 | 2023-05-16 | Walmart Apollo, Llc | System and method for electronic text classification |
US11276040B2 (en) | 2018-12-14 | 2022-03-15 | International Business Machines Corporation | Identifying complaints from messages |
US10984388B2 (en) * | 2018-12-14 | 2021-04-20 | International Business Machines Corporation | Identifying complaints from messages |
US20200193380A1 (en) * | 2018-12-14 | 2020-06-18 | International Business Machines Corporation | Detection and categorization of electronic communications |
Also Published As
Publication number | Publication date |
---|---|
TW201033823A (en) | 2010-09-16 |
WO2010066616A1 (en) | 2010-06-17 |
US20100145940A1 (en) | 2010-06-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US8606815B2 (en) | Systems and methods for analyzing electronic text | |
Günther et al. | Word counts and topic models: Automated text analysis methods for digital journalism research | |
Tang et al. | A survey on sentiment detection of reviews | |
US8356025B2 (en) | Systems and methods for detecting sentiment-based topics | |
US11023478B2 (en) | Determining temporal categories for a domain of content for natural language processing | |
US7685091B2 (en) | System and method for online information analysis | |
Petersen et al. | A machine learning approach to reading level assessment | |
Li et al. | Application of a clustering method on sentiment analysis | |
Abbasi et al. | Selecting attributes for sentiment classification using feature relation networks | |
Gamon et al. | Pulse: Mining customer opinions from free text | |
US6836768B1 (en) | Method and apparatus for improved information representation | |
US9836455B2 (en) | Apparatus, method and computer-accessible medium for explaining classifications of documents | |
US7882115B2 (en) | Method and apparatus for improved information representation | |
US8543565B2 (en) | System and method using a discriminative learning approach for question answering | |
US20080040339A1 (en) | Learning question paraphrases from log data | |
Durant et al. | Predicting the political sentiment of web log posts using supervised machine learning techniques coupled with feature selection | |
US20150356170A1 (en) | Time-Based Optimization of Answer Generation in a Question and Answer System | |
Chelaru et al. | Analyzing, detecting, and exploiting sentiment in web queries | |
Ferreira et al. | A comparative study of feature extraction algorithms in customer reviews | |
Zhao et al. | WTL-CNN: A news text classification method of convolutional neural network based on weighted word embedding | |
Woloszyn et al. | Distrustrank: Spotting false news domains | |
Lei et al. | Examining research topics with a dependency-based noun phrase extraction method: a case in accounting | |
Wang et al. | Answer selection and expert finding in community question answering services: A question answering promoter | |
Li et al. | Seeded-btm: enabling biterm topic model with seeds for product aspect mining | |
Saggion et al. | Extracting Opinions and Facts for Business Intelligence. |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AS | Assignment |
Owner name: INTERNATIONAL BUSINESS MACHINES CORPORATION,NEW YO Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:CHEN, YING;PROCTOR, LARRY;SPANGLER, WILLIAM SCOTT;SIGNING DATES FROM 20090114 TO 20090123;REEL/FRAME:022245/0491 Owner name: INTERNATIONAL BUSINESS MACHINES CORPORATION, NEW Y Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:CHEN, YING;PROCTOR, LARRY;SPANGLER, WILLIAM SCOTT;SIGNING DATES FROM 20090114 TO 20090123;REEL/FRAME:022245/0491 |
|
STCF | Information on status: patent grant |
Free format text: PATENTED CASE |
|
FPAY | Fee payment |
Year of fee payment: 4 |
|
FEPP | Fee payment procedure |
Free format text: MAINTENANCE FEE REMINDER MAILED (ORIGINAL EVENT CODE: REM.); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY |
|
AS | Assignment |
Owner name: DOORDASH, INC., CALIFORNIA Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:INTERNATIONAL BUSINESS MACHINES CORPORATION;REEL/FRAME:057826/0939 Effective date: 20211012 |
|
LAPS | Lapse for failure to pay maintenance fees |
Free format text: PATENT EXPIRED FOR FAILURE TO PAY MAINTENANCE FEES (ORIGINAL EVENT CODE: EXP.); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY |
|
STCH | Information on status: patent discontinuation |
Free format text: PATENT EXPIRED DUE TO NONPAYMENT OF MAINTENANCE FEES UNDER 37 CFR 1.362 |
|
FP | Lapsed due to failure to pay maintenance fee |
Effective date: 20211210 |